Analysis of naphthenic acid mixtures as pentafluorobenzyl derivatives by gas chromatography-electron impact mass spectrometry

Talanta 162 (2017) 440–452

Contents lists available at ScienceDirect

Talanta journal homepage: www.elsevier.com/locate/talanta

Analysis of naphthenic acid mixtures as pentafluorobenzyl derivatives by gas chromatography-electron impact mass spectrometry

crossmark

Juan Manuel Gutierrez-Villagomeza, Juan Vázquez-Martínezb, Enrique Ramírez-Chávezb, ⁎ Jorge Molina-Torresb, Vance L. Trudeaua, a b

Department of Biology, University of Ottawa, Ottawa, Ontario, Canada K1N 6N5 Departamento de Biotecnología y Bioquímica, CINVESTAV Unidad Irapuato, Guanajuato, Mexico

A R T I C L E I N F O

A BS T RAC T

Keywords: Naphthenic acids Derivatization PFBBr Mixtures GC/EIMS Oil

In this study, we report for the first time the efficiency of pentafluorobenzyl bromide (PFBBr) for naphthenic acid (NA) mixtures derivatization, and the comparison in the optimal conditions to the most common NAs derivatization reagents, BF3/MeOH and N-(t-butyldimethylsilyl)-N-methyltrifluoroacetamide (MTBSTFA). Naphthenic acids are carboxylic acid mixtures of petrochemical origin. These compounds are important for the oil industry because of their corrosive properties, which can damage oil distillation infrastructure. Moreover, NAs are commercially used in a wide range of products such as paint and ink driers, wood and fabric preservatives, fuel additives, emulsifiers, and surfactants. Naphthenic acids have also been found in sediments after major oils spills in the United States and South Korea. Furthermore, the toxicity of the oil sands processaffected water (OSPW), product of the oil sands extraction activities in Canada's oil sands, has largely been attributed to NAs. One of the main challenges for the chromatographic analysis of these mixtures is the resolution of the components. The derivatization optimization was achieved using surface response analysis with molar ratio and time as factors for derivatization signal yield. After gas chromatography-electron impact mass spectrometry (GC/EIMS) analysis of a mixture of NA standards, it was found that the signal produced by PFB-derivatives was 2.3 and 1.4 times higher than the signal produced by methylated and MTBS-derivatives, respectively. The pentafluorobenzyl derivatives have a characteristic fragment ion at 181m/z that is diagnostic for the differentiation of carboxylic and non-carboxylic acid components within mixtures. In the analysis of a Sigma and a Merichem derivatized oil extract NA mixtures, it was found that some peaks lack the characteristic fragment ion; therefore they are not carboxylic acids. Open column chromatography was used to obtain a hexane and a methanol fraction of the Sigma and Merichem mixtures. The components in the hexane fraction, presumably hydrocarbons that did not react with PFBBr were ~7% by weight. The effectiveness of PFBBr was confirmed when the two NA oil extracts were spiked with 8 distinct NA standards and identified by GC/EIMS in the methanol fraction. Here, we also report retention indices of the methyl, MTBS and PFB derivatives of these 8 NAs. The use of PFBBr increases sensitivity, chromatographic resolution, and identification accuracy for the analysis of standards and mixtures of NAs compared to MTBSTFA and BF3/MeOH. This methodology will have wide applications in the elucidation of NA mixtures.

1. Introduction The IUPAC defines naphthenic acids (NAs) as “acids chiefly monocarboxylic, derived from naphthene” [1]. However, the oil industry has adopted the term NAs to encompass all the carboxylic acids present in crude oil which includes a mixture of alkyl-substituted cyclic and aliphatic carboxylic acids (Fig. 1) [2]. Not all of these are naphthalene derivatives, thus the industry definition will be used here. NAs are represented by the general formula: CnH2n−zO2, where n

⁎

indicates the carbon number, z specifies the deficiency of hydrogen from the formation of rings and groups the NAs in a homologous series (Fig. 1) [2]. Although, NAs correspond to 0–3% of oil by weight [2] they are important for the oil industry because of their corrosive properties [3,4] which can damage oil distillation towers [5]. Moreover, NAs are also used commercially in a wide range of products such as paint and ink driers, wood and fabric preservatives, fuel additives, emulsifiers, and surfactants. NAs have been found in sediments after major oil

Corresponding author. E-mail address: [email protected] (V.L. Trudeau).

http://dx.doi.org/10.1016/j.talanta.2016.10.057 Received 24 July 2016; Received in revised form 12 October 2016; Accepted 13 October 2016 Available online 14 October 2016 0039-9140/ © 2016 Elsevier B.V. All rights reserved.

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

HPLC-grade (Sigma-Aldrich Co. Ltd, St. Louis, United States). BF3/ MeOH (14% w/v), MTBSTFA, PFBBr were used as purchased (SigmaAldrich Chemie, Steinheim, Germany). NA standards were selected to cover a variety of chemical structures and molecular weights. A C7 to C40 alkane calibration standard (Sigma-Aldrich Chemie, Steinheim, Germany) was used to calculate the retention indices (Ri). 2.2. Derivatization For the 3 derivatization reactions, an equimolar solution of all of the standards was prepared in methanol. From this solution, a defined volume containing 1 µmol of each standard was transferred to vials and dried under nitrogen flux, to later proceed with the derivatizations. Sigma and Merichem extract stock solutions were prepared in methanol. For these derivatizations, an average molecular weight of 184 amu was assumed based on preliminary experiments. From these solutions, a defined volume containing 5 µmol was transferred to vials and dry under nitrogen flux to later proceed with the derivatizations. To obtain the methyl esters of the NAs, the derivatization was performed by adding BF3/MeOH in different volumes depending on the reaction molar ratio ( Table 2) and the reaction was adjusted to 500 µL with acetonitrile as organic solvent. Samples were then heated to 60 °C and mixed at 1000 rpm. To recover the methyl esters, 500 µL of hexane and 500 µL of water were added and the resulting mix was vortexed. The mix was then centrifuged for 2 min at 6000 rpm to separate the two phases. An aliquot of the hexane phase was taken and mixed with isooctane to obtain a final concentration of 0.3 mM for each methylated standard and 6 µg µL−1 for the oil extracted NA mixtures, assuming 100% methyl esterification. The silyl derivatives of the mixture of NA standards and the oil extracted NA mixtures were obtained with MTBSTFA ( > 97% purity), and 20 µL of pyridine (99% purity) in each reaction as a catalyst. MTBSTFA was added according to Table 2. The reaction volume was adjusted to 200 µL with isooctane as organic solvent. Samples were then heated to 80 °C and mixed at 1000 rpm. Once the reaction finalized, the solvent, the MTBSTFA, and the pyridine excess were evaporated to dryness under nitrogen flux, and the residue was redissolved in isooctane to a final concentration of 0.3 mM for each standard and a concentration of 6 µg µL−1 for the oil extracted NA mixtures. In addition to BF3/MeOH and MTBSTFA, we propose pentafluorobenzyl bromide (PFBBr) for the derivatization and analysis of NAs. This idea emerged from the necessity to improve NA resolution during the chromatography phase to analyze the components within mixtures. The pentafluorobenzyl group is a relatively stable ion that herein is found as characteristic ion for PFB-carboxylic acid derivatives. The derivatization reaction with PFBBr can be selective for NAs and other carboxylic acids if the reaction is done using a weak base [38,39]. The weak base used for these reactions was N,N-diisopropylethylamine. PFBBr was added according to Table 2 and 50 µL of N,N-diisopropylethylamine were added in each reaction as a catalyst and in order to make the reaction selective to carboxylic acids. The reaction volume was adjusted to 200 µL with chloroform as organic solvent. The samples were then heated to 60 °C and mixed at 1000 rpm. At the end of the reaction the solvent, the weak base, and the PFBBr excess were evaporated to dryness under nitrogen flux, the residue was redissolved in a mixture of chloroform/methanol 1:1 to a final concentration of 0.3 mM for each standard and a concentration of 6 µg µL−1 for the oil extracted NA mixtures. Additionally, estradiol-17β (SigmaAldrich, ≥98%) was used as negative control for the selective derivatization of carboxylic acids with PFBBr using a weak base.

Fig. 1. Conventional NAs structures grouped in z families. R=alkyl group.

spills in the United States [6,7] and South Korea [8]. The toxicity of the oil process-affected water (OSPW), product of the oil sands extraction activities in Canada's oil sands, has largely been attributed to the presence of NAs [9,10]. The chemical composition of NA mixtures is highly diverse and may also contain non-classical NAs that do not fit the general formula of NAs [11], nitrogen or sulfur containing compounds [12], alcohols, ketones, and ethers [13]. The analysis of such mixtures has been widely explored [14–19], however, the main challenge for their characterization is the resolution of the components in the chromatographic phase. In this regard, the use of derivatization techniques to improve the chromatographic characteristics of the mixtures of NAs by GC–MS has been reported previously. Despite the problems summarized by Walker Christie [20] and the superior results offered by silylation methods [21], the alkylation reaction of NAs with BF3/MeOH remains widely used [22–28]. In the methylation reaction, the labile hydrogen of the carboxylic group is substituted with a methyl group [29]. Silylation is also commonly used for the analysis of NA mixtures [30–35], where in the case of N-(t-butyldimethylsilyl)-N-methyltrifluoroacetamide (MTBSTFA) the labile hydrogens of carboxylic acids, alcohols, phenols and amines, among others, are replaced with a t-butyldimethylsilyl (tBDMS) group [29]. This report describes the optimization and comparison of three derivatization reagents in a mixture of NA standards and two oil extracted NA mixtures. The comparison was performed using optimal conditions of molar ratio (derivatization reagent/NA) and time. The derivatization reagents were BF3/MeOH, MTBSTFA, and an alkylation reagent not previously reported for the analysis of NA mixtures: alphabromo-2,3,4,5,6-pentafluorotoluene (pentafluorobenzyl bromide or PFBBr). In the reaction with PFBBr, the labile hydrogen of the carboxylic group is substituted with a pentafluorobenzyl group [29]. PFBBr can be used for the derivatization of phenols, thiols and carboxylic acids [36,37]. However, the reaction is selective to carboxylic acids in the presence of a weak base, e.g., KOAc, KHCO3, KCNO or organic bases [38,39]. PFBBr has been used successfully for the analysis of mixtures of compounds with carboxylic groups, for example, carboxylic acids and phenols in air [40], hydroxy acids in wine [41] and fatty acids in blood plasma [42]. In this work, we report that PFBBr derivatization increases sensitivity, chromatographic resolution and NAs identification accuracy for NA mixtures analysis by GC/EIMS.

2. Experimental 2.1. Reagents and chemicals Eight single NA standards were purchased from two suppliers, as describe in Table 1. One of the NA mixtures was purchased from Sigma-Aldrich Co. Ltd. (St. Louis, United States), with lot number BCBC9959V. The Merichem NA mixture was a gift from Dr. John Headley (Environment Canada, Saskatoon, SK). All solvents were

2.3. Response surface analysis In order to find the optimum conditions for the three derivatization reactions, response surface methodology (RSM) with center composite 441

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

Table 1 Individual NA standards for derivatization evaluation. Compound

Name

Molecular weight

Basic structure description

z family

Purity (%)

Supplier

1 2 3 4 5 6 7 8

Hexanoic acid Cyclopentanecarboxylic acid Cyclohexanecarboxylic acid Cyclohexanepentanoic acid 1-adamantanecarboxylic acid 1-adamantaneacetic acid 3-Methyl-1-adamantaneacetic acid Dehydroabietic acid

116.16 114.14 128.17 184.28 180.24 194.27 208.3 300.44

Aliphatic Monocyclic Monocyclic Monocyclic Adamantane Adamantane Adamantane Aromatic, tricyclic

0 2 2 2 6 6 6 12

99 99 ≥98 98 99 98 99 ≥99

Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich Sigma-Aldrich CanSyn Chem. Corp.

Table 2 Factors and results for the derivatization of a mixture of NA standards.

1 2 3 4 5 6 7 8 9 10

MTBSTFA Time (min)

Signal yield

Molar ratio

Time (min)

Signal yield

Molar ratio

Time (min)

Signal yield

250 250 250 750 750 750 750 1250 1250 1250

10 30 50 10 30 30 50 10 30 50

3.2 5.9 7.5 7.1 10.8 11.8 13.4 12.0 14.4 14.0

25 25 25 50 50 50 50 75 75 75

10 30 50 10 30 30 50 10 30 50

11.8 11.0 9.2 9.4 10.7 10.6 9.2 9.5 9.9 8.7

10 10 10 30 30 30 30 50 50 50

10 30 50 10 30 30 50 10 30 50

6.7 9.8 10.2 10.8 10.3 10.8 10.5 11.1 10.1 9.7

b)

50 Signal yield Yield < 4.5 4.5 – 6.0 6.0 – 7.5 7.5 – 9.0 9.0 – 10.5 10.5 – 12.0 12.0 – 13.5 13.5 – 15.0 > 15.0

Time (min)

40

30

20

10

400

PFBBr

Molar ratio

600

800

1000 1200

Molar ratio

c)

50 Signal yield Yield < 9.0 9.0 – 9.4 9.4 – 9.8 9.8 – 10.2 10.2 – 10.6 10.6 – 11.0 11.0 – 11.4 > 11.4

40

Time (min)

a)

BF3/MeOH

30

20

10

30

40

50

60

70

50 SignalYield yield < 7.5 7.5 – 8.0 8.0 – 8.5 8.5 – 9.0 9.0 – 9.5 9.5 – 10.0 10.0 – 10.5 10.5 – 11.0 > 11.0

40

Time (min)

Experiment

30

20

10 10

20

Molar ratio

30

40

50

Molar ratio

Fig. 2. Contour plots of derivatization signal yield vs molar ratio and reaction time of the three derivatization reagents and the mixture of NA standards. BF3/MeOH (a; R2=94.16), for MTBSTFA (b; R2=86.44) and for PFBBr (c; R2=85.49).

face design (CCFD) was used to evaluate the effect of the molar ratio of each derivatization reagent/NA and the reaction time. Each factor had three levels −1, 0 and 1. The conditions of the reactions are shown in Table 2. This methodology has been widely used in analytical chemistry [43]. The derivatization performance was evaluated according to the signal yielded or area under the peak by the derivatized standards measured by the GC/EIMS and analyzed in the chromatograms using the methodology previously reported by Shepherd et al. [21]. Briefly, 10 experiments were carried out per derivatization to obtain 10 chromatograms with 8 peaks in each chromatogram. The signal produced by these 80 peaks correspond to 100% of the signal. The percentage of the total signal that is represented by each out of the 80 peaks was obtained with the formula NAsignal-yield=(peak area)/ (total area)*100. The signal produced by the derivatized standards according to the conditions set for each of the 10 experiments was calculated using the formula Experiment signal yield=∑NAsignalyield. The same principle was applied for the comparison of the 3 different derivatizations. Briefly, the mixture of standards was derivatized under optimized conditions as determined by RSM analysis. Three chromatograms were obtained with 8 peaks in each chromatogram. The area of these 24 peaks correspond to 100% of the signal produced by the derivatives. The percentage of the signal that is represented by each peak was calculated using the formula NAsignal-

yield=(peak

area)/(total area)*100 and a summation was used to know the percentage of the whole signal that is represented by each type of derivatization. 2.4. Open column chromatography The chromatographic columns were prepared by loading the 0.7 cm×10 cm glass columns with silica gel 60 (Merck KGaA, Darmstadt, Germany) suspended in cold hexane so that the column packing was ~8 cm long. Each column was flushed with 3 mL of cold hexane, then loaded with the respective oil extracted NA mixture (0.1g) as a “neat oil” and it was allowed to pass through the column by gravity. Once the oil extracted NA mixture was fully loaded, hexane was added until 2 mL was recovered from the column to collect the hexane fraction. This was followed by 1 mL of ethyl acetate and 2 mL of methanol until all the fraction passed through the column. This process was done for the 2 oil extracted NA mixtures. Additionally, the 2 oil extracted NA mixtures were spiked with 1.3 μmol of each of the 8 NA standards and separated by open column chromatography to confirm the separation of carboxylic acids (methanol fraction) from hexane fraction in the column. The 2 fractions obtained in each of the column experiments (methanol and hexane fraction Sigma extract, methanol and hexane fraction Merichem extract, methanol and hexane fraction

442

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

a) 6E+06 6E+06

5E+06

R6-7=0.7

R4-5=2.1

Abundance

4E+06

7 6

4E+06

2E+06

5

3E+06 2E+06

0E+00 20.8

21

21.2 21.4 21.6 21.8

22

22.2 22.4 22.6

4

1E+06

1 2

3 8

0E+00

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Retention Time (min) b) 6E+06

6E+06

R4-5=0.8

R7-6=1.0

4E+06

6 57

Abundance

5E+06 4E+06

2E+06 0E+00 25.3 25.5 25.7 25.9 26.1 26.3 26.5 26.7 26.9 27.1

4

8

3E+06

3

2E+06

1 2

1E+06 0E+00

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Retention Time (min) c) 6E+06

576

Abundance

5E+06

4

6E+06

R4-5=3.4

R7-6=0.9

4E+06 2E+06

4E+06

0E+00 26.7 26.9 27.1 27.3 27.5 27.7 27.9 28.1 28.3 28.5

3

3E+06

2

8

1

2E+06 1E+06 0E+00

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45

Retention Time (min) Fig. 3. GC/EIMS TICs for a derivatized mixture of NA standards. (a) methylated NAs, (b) MTBS-NAs and (c) PFB-NAs. Numbers correspond to NAs derivatives following the codes of Table 1. R represents the resolution of two adjacent peaks calculated according to the IUPAC resolution equation for chromatography.

Sigma extract spiked with standards, methanol and hexane fraction Merichem extract spiked with standards) were dried under nitrogen flux, weighted, later derivatized with PFBBr and analyzed by GC-EIMS.

Table 3 Comparison of NA signal yields for different chemical derivatizations applied to a mixture of NA standards. Derivatization reagent

∑ NA signal yield percent

Average NA signal yield percent

BF3/MeOH MTBSTFA PFBBr

20.1 33.5 46.3

2.5 4.2 5.8

2.5. Instrumentation and software A gas chromatograph (Agilent Technologies model 7890A GC System) coupled with an electron impact ionization mass spectrometer (Hewlett Packard model 5973 Mass Selective Detector) was used for 443

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

Fig. 4. Mass spectra of methylated NAs (a)–(d), MTBS-NAs (e)–(h), and PFB-NAs (i)–(l), and their corresponding structures.

444

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

Fig. 4. (continued)

tives.

the analysis. The data obtained by the GC-EIMS was collected with the software MassHunter Workstation version B.06.00 (Agilent Technologies, Inc.). Agilent MassHunter Qualitative Analysis version B.06.00 was used for the ion distribution analysis of the commercial mixtures. The software Automated Mass Spectral Deconvolution and Identification System “AMDIS” (http://www.amdis.net/) was used for the determination of the retention time and mass spectrum for each component of the chromatograms of the NA standards and oil extracted NA mixtures. MSD ChemStation Data Analysis version E. 02.01.1177 was used to calculate the linearity of the response of the derivatized standards. Minitab 17 was used for the analysis of the surface response analysis. The conditions and parameters of the chromatography phase were adjusted to obtain the best mixture components resolution with the lowest interference of the noise signal. All the parameters of the MS were optimized to obtain the best signal/noise ratio of the analytes. Pulsed splitless injection (1 µL) was used. The injector temperature was set to 250 °C. The components separation was performed in a capillary column Zebron ZB-1MS (60 m×320 µm×1 µm) and helium used as carrier gas at a constant flow rate of 1 mL min−1. The GC oven program began at an initial temperature of 50 °C, held for the first minute, and then was increased at a rate of 10 °C min−1 to a final temperature of 300 °C, held for 35 min. The transfer line temperature was set at 260 °C. Electron impact mass spectra were obtained at 70 eV of electron energy. Measurements were performed in SCAN mode with m/z range set to 40–550. The ion source and quadrupole analyzer temperature were 230 °C and 150 °C respectively and operated at 2.9 scans per second. A different solvent delay was selected for each type of derivatization to avoid damage to the MS filament, without disrupting the measurement of low molecular weight NAs. The solvent delay for BF3/MeOH was 11 min and 15 min for MTBSTFA and PFBBr deriva-

2.6. Data processing The Ri of the derivatives from the NA standards were calculated by alkane linear retention indices (C7 to C40) following the methodology of Sun and Stremple [44]. Using the optimal conditions for the 3 derivatizations, a concentration series of the 3 derivatized mixtures of NA standards ranging from 0.0006 to 0.3 mM were analyzed in order to determine linearity, limits of detection and limits of quantification. Calibration curves were generated to assess the linearity of the GC/EIMS measurements using ChemStation. The linear ranges varied from 0.001 to 0.3 mM. Limits of detection and quantification were calculated using AMDIS for each of the eight standards on the basis of a signal to noise ratio of 3 and 10, respectively [50]. To generate the ion distribution of the oil extract mixtures, a custom database of possible NA formulas was created spanning n=4–27 and z=0–10 starting from the formula of classical NAs (CnH2n−zO2), using the methodology reported by Clemente et al. [45]. Only peaks with areas greater than 1% compared to the greatest peak were considered for analysis. The resolution of adjacent peaks was calculated according to the IUPAC formula for resolution in GC [1]. 3. Results and discussion 3.1. Response surface methodology The relationship between the derivatization signal yield, the molar ratio and derivatization time was determined using RSM. Ten experiments were carried out per derivatization and the results are shown in Table 2. Analysis of variance (ANOVA) with alpha (α) level of 0.05 showed the significance of all of the quadratic models. After a stepwise 445

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

Table 4 Chromatographic and mass spectral properties of NA derivatives by GC/EIMS using BF3/MeOH, MTBSTFA, and PFBBr. Relative abundances to maximum are given in parentheses (base 100). Methylated NAs M+

Analyte

Rt

Ri

Fragmentation pattern (m/z) and relative abundance (%)

Hexanoic acid, methyl ester Cyclopentanecarboxylic acid, methyl ester Cyclohexanecarboxylic acid, methyl ester Cyclohexanepentanoic acid, methyl ester 1-Adamantanecarboxylic acid, methyl ester 1-adamantaneacetic acid, methyl ester 3-Methyl-1adamantaneacetic acid, methyl ester Dehydroabietic acid, methyl ester

12.126 12.922

906 950

101 (8.8) 128 (6.8)

99 (19.3) 100 (20.3)

87 (34.8) 97 (15.1)

74 (99.9) 87 (99.9)

71 (9.3) 69 (51.6)

59 (27.8) 68 (12.8)

55 (18.6) 67 (16.3)

43 (52.9) 59 (8.1)

42 (18.1) 55 (17.3)

41 (27.1) 41 (46.1)

130 128

14.777

1054

142 (22.8)

110 (174)

87 (68.8)

83 (61.2)

82 (20.9)

74 (27.5)

67 (21.5)

59 (17.6)

55 (99.9)

41 (49.6)

142

21.031

1474

148 (17.1)

122 (37.0)

87 (75.5)

83 (21.4)

81 (13.4)

74 (99.9)

67 (20.1)

59 (19.8)

55 (81.1)

41 (49.3)

198

21.158

1483

194 (9.5)

136 (10.9)

135 (99.9)

107 (9.8)

93 (20.4)

91 (13.4)

79 (26.2)

77 (13.3)

67 (7.5)

41 (8.4)

194

22.307

1575

136 (11.0)

135 (99.9)

107 (6.3)

93 (12.7)

92 (8.0)

91 (17.0)

79 (19.5)

77 (9.6)

67 (6.4)

41 (7.6)

208

22.353

1579

150 (12.0)

149 (99.9)

133 (7.6)

107 (12.2)

105 (14.9)

93 (28.6)

91 (21.3)

79 (92)

77 (10.2)

41 (9.4)

222

31.912

2379

281 (12.6)

240 (20.5)

239 (99.9)

208 (10.7)

141 (10.8)

133 (9.6)

128 (10.0)

59 (11.0)

43 (12.0)

41 (11.0)

314

Rt 18.479

Ri 1288

Fragmentation pattern (m/z) and relative abundance (%) 174 (11.0) 173 (84.0) 131 (13.9) 76 (7.1) 75 (99.9)

73 (15.4)

47 (6.3)

45 (7.0)

43 (8.9)

41 (13.6)

M+ ND

19.233

1341

172 (13.4)

171 (99.9)

76 (72)

75 (98.4)

73 (18.4)

69 (12.8)

57 (8.0)

47 (6.7)

45 (7.7)

41 (24.0)

ND

20.689

1448

186 (14.7)

185 (99.9)

83 (7.4)

76 (6.5)

75 (89.3)

73 (16.3)

57 (8.6)

55 (14.8)

45 (6.2)

41 (17.9)

ND

25.645

1870

242 (19.9)

241 (99.9)

131 (10.5)

129 (30.0)

117 (11.8)

81 (10.6)

75 (74.6)

73 (17.4)

55 (31.0)

41 (17.4)

298

25.692

1874

238 (19.6)

237 (99.9)

135 (45.6)

93 (10.1)

91 (7.4)

79 (13.7)

77 (7.7)

75 (19.7)

73 (11.4)

41 (10.9)

ND

26.753

1975

266 (21.6)

265 (99.9)

150 (11.6)

149 (96.2)

107 (11.5)

93 (29.3)

91 (14.0)

75 (37.0)

73 (18.9)

41 (13.3)

ND

26.811

1981

252 (19.8)

251 (96.5)

136 (11.1)

135 (99.9)

93 (14.3)

91 (11.0)

79 (17.7)

75 (36.2)

73 (16.6)

41 (12.5)

ND

38.024

2670

358 (29.2)

357 (99.9)

239 (35.2)

173 (14.3)

171 (22.6)

143 (21.8)

117 (28.8)

115 (14.5)

75 (40.5)

73 (38.8)

414

Rt 20.408

Ri 1427

Fragmentation pattern (m/z) and relative abundance (%) 181 (99.9) 161 (10.1) 99 (8.9) 97 (20.4) 73 (9.7)

71 (8.0)

69 (32.2)

55 (11.4)

43 (26.9)

41 (17.3)

M+ 296

21.208

1487

182 (7.7)

181 (96.4)

161 (14.7)

113 (15.0)

97 (10.0)

95 (31.9)

69 (99.9)

67 (45.9)

55 (8.8)

41 (64.4)

294

22.531

1593

181 (99.9)

161 (15.5)

127 (12.9)

109 (40.4)

83 (56.1)

81 (82.7)

67 (8.8)

55 (98.3)

53 (8.6)

41 (42.6)

308

27.025

2001

181 (99.9)

165 (38.2)

147 (19.0)

123 (10.9)

83 (30.6)

81 (29.3)

69 (14.2)

67 (19.8)

55 (74.7)

41 (35.3)

36

27.23

2020

181 (33.0)

136 (11.2)

135 (99.9)

107 (9.0)

93 (18.6)

91 (9.8)

79 (21.3)

77 (9.9)

67 (6.8)

41 (6.0)

360

28.244

2110

181 (54.5)

150 (13.4)

149 (99.9)

147 (12.9)

107 (10.5)

105 (12.4)

93 (24.4)

91 (17.9)

79 (9.9)

77 (8.4)

388

28.302

2115

181 (50.0)

136 (11.3)

135 (99.9)

133 (18.6)

93 (19.5)

91 (22.3)

79 (22.2)

77 (11.5)

67 (9.5)

41 (9.5)

374

41.707

2794

240 (20.1)

239 (99.9)

181 (62.9)

155 (9.7)

143 (9.0)

141 (10.9)

131 (8.6)

129 (10.1)

128 (9.4)

43 (11.7)

480

MTBS-NAs Analyte Hexanoic acid, tertbutyldimethylsilyl ester Cyclopentanecarboxylic acid, tertbutyldimethylsilyl ester Cyclohexanecarboxylic acid, tert-butyldimethylsilyl ester Cyclohexanepentanoic acid, tert-butyldimethylsilyl ester 1-Adamantanecarboxylic acid, tertbutyldimethylsilyl ester 3-Methyl-1adamantaneacetic acid, tert-butyldimethylsilyl ester 1-Adamantaneacetic acid, tert-butyldimethylsilyl ester Dehydroabietic acid, tertbutyldimethylsilyl ester PFB-NAs Analyte Hexanoic acid, pentafluorobenzyl ester Cyclopentanecarboxylic acid, pentafluorobenzyl ester Cyclohexanecarboxylic acid, pentafluorobenzyl ester Cyclohexanepentanoic acid, pentafluorobenzyl ester 1-Adamantanecarboxylic acid, pentafluorobenzyl ester 3-Methyl-1adamantaneacetic acid, pentafluorobenzyl ester 1-Adamantaneacetic acid, pentafluorobenzyl ester Dehydroabietic acid, pentafluorobenzyl ester ND, non-detected

z=4.09+0.2769x+0.1069y−0.00250x2−0.003062xy (p=0.025) for PFBBr (z=signal yield, x=molar ratio and y=time). The unexplained variance for each model is 5.84%, 13.56%, and 14.51%, for BF3/MeOH, MTBSTFA, and PFBBr, respectively. The significance of each of the

selection of terms where the terms with alpha level greater than 0.15 were removed, the equations obtained were: z=−1.19+0.01558x +0.1050y−0.000005x2 (p < 0.001) for BF3/MeOH, z=12.06−0.0534x +0.0619y−0.002293y2+0.000913xy (p=0.021) for MTBSTFA and 446

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

Table 5 Comparison of limits of detection (LOD) and limits of quantification (LOQ) of the eight standards measured by GC/EIMS using BF3/MeOH, MTBSTFA, and PFBBr as derivatization reagents. Linear equations and coefficients of determination (r2) are also described. BF3 MeOH

MTBSTFA

PFBBr

Standard

Linear equation

r2

LQ (mM)

LD (mM)

Linear equation

r2

LQ (mM)

LD (mM)

Linear equation

r2

LQ (mM)

LD (mM)

Hexanoic acid

y=4E+07x −2E+06 y=4E+07x −331551 y=6E+07x −478784 y=2E+08x −2E+07 y=3E+08x −2E+06 y=3E+08x −3E+06 y=3E+08x −2E+06 y=6E+06x −435680

0.9957

0.0134

0.0078

0.9941

0.0044

0.0016

0.007

0.0042

0.0055

0.0034

0.9951

0.00205

0.00065

0.9974

0.0026

0.0005

0.997

0.0022

0.0001

0.9959

0.0016

0.0002

0.9974

0.0024

0.001

0.978

0.0178

0.0136

0.9818

0.0065

0.0037

0.9963

0.0042

0.0021

0.9952

0.0012

0.0005

0.9975

0.0008

0.0001

0.9977

0.0008

0.00038

0.9878

0.0015

0.0001

0.9968

0.00197

0.00057

0.999

0.0007

0.0001

0.9922

0.00194

0.00054

0.9937

0.0018

0.0004

0.9944

0.0017

0.0003

0.9834

0.0393

0.0134

0.9932

0.0057

0.0043

y=1E+08x −2E+06 y=2E+08x −1E+06 y=2E+08x −2E+06 y=4E+08x −9E+06 y=4E+08x −651360 y=4E+08x −623204 y=5E+08x −5E+06 y=5E+08x −6E+06

0.9945

0.9987

y=5E+07x −418382 y=6E+07x −472855 y=1E+08x −1E+06 y=3E+08x −1E+07 y=3E+08x −1E+06 y=3E+08x −2E+06 y=4E+08x −5E+06 y=3E+08x −3E+06

0.9938

0.003

0.0002

Cyclopentanecaboxylic acid Cyclohexanecarboxylic acid Cyclohexanepentanoic acid 1-Adamantanecarboxylic acid 1-Adamantaneacetic acid 3-Methyl-1-adamantaneacetic acid Dehydroabietic acid

y: peak area; x: concentration (mM).

The results showed that PFBBr is the most effective among the three derivatization reagents when considering their derivatization signal yields (Fig. 3, Table 3). The signal produced by PFB-derivatives is 2.3 times higher than the signal produced from using the methylation reagent and 1.4 times higher to the silylation reagent. This could be due to the addition of 180 amu to the model NAs, increasing the sensitivity of the GC/EIMS [46]. PFBBr might also have a greater chemical derivatization yield, an increase in the volatility of the PFBderivatives or increase in ionization efficiency for PFBBr derivatives. However, this is speculative. Furthermore, MTBSTFA has higher signal yields compared to BF3/MeOH (Table 3). The span of the retention times for the first and last standards eluted for methylated NAs, MTBS-NAs and PFB-NAs were 19.8, 19.5, and 21.3 min, respectively. This is a result of the increase of molecular weights according to the derivatization reagent employed. It is important to note that reaction with BF3/MeOH and MTBSTFA give an aliphatic group, however, PFBBr gives a halogenated-aromatic group, altering the interaction of the compounds with the column. On average PFB-derivatives show 54% and 140% more chromatographic resolution compared to methylated and silyl derivatives, respectively (Fig. 3). Due to the complexity of the mixtures, only a few NAs have been clearly identified previously [24]. However, the use of PFBBr as derivatization reagent for NA mixtures may improve the analysis and identification of NAs, since PFBBr increases the span of elution time and resolution of NAs with 6–20 carbons. Furthermore, the preparation of PFB-derivatives is more efficient since the derivatization reaction with PFBBr takes 10 min producing a higher signal. In comparison the methylation and silylation reactions take 50 and 18.5 min, respectively, producing a lower derivatization signal yield (Table 3, Fig. 3).

terms of the models is shown in Table S1. The cross validation analysis indicated the predicted R2 for each of the models of 81.9%, 10.6%, and 3.7%, for BF3/MeOH, MTBSTFA, and PFBBr, respectively. Thus, there are likely other factors affecting the derivatization of NAs with MTBSTFA and PFBBr. The predicted R2 value of the BF3/MeOH model indicates a greater predictive ability than the MTBSTFA and PFBBr models. The contour and 3D surface plots for the model of each derivatization reagent show the responses of the molar ratio of derivatization reagent/NA and the time of reaction (Figs. 2 and S1). The plots (Figs. 2a and S1a) show a maximum derivatization signal yield for the conditions here tested for BF3/MeOH at 1250 M ratio and 50 min of reaction. The optimal conditions for MTBSTFA (Figs. 2b, and S1b) were observed at 25 M ratio and 18.5 min. PFBBr plots (Figs. 2c and S1c) show optimal conditions at 49 M ratio and 10 min. It was observed that there is an increase in the derivatization signal yield when the molar ratio and the time of reaction increase with BF3/MeOH (Figs. 2a and S1a). The derivatization using BF3/MeOH is disadvantageous compared to MTBSTFA and PFBBr because it takes more than 50 min for the reaction to reach maximum signal yield and requires a large molar excess of reagents. With MTBSTFA, the area for optimal derivatization was ~25–30 M ratio and 10–25 min of reaction, however beyond the optimal conditions, the derivatization signal yield decreased (Figs. 2b and S1b). For PFBBr derivatization there is a wide area of optimal signal yield with the interaction of the time and the molar ratio (Figs. 2c and S1a). The optimal conditions were selected to compare the results of the three derivatization reagents according to their derivatization signal yield.

3.2. GC analysis of mixtures of standards 3.3. Mass spectral analysis of NA standards Fig. 3 shows the GC/EIMS total ion chromatograms (TIC) of the derivatized NA standards mixture as listed in Table 1. In these chromatograms, it is possible to observe the elution order and retention time for NA standards in relation to the derivatization reagent used. The methylated-NAs had shorter retention times, followed by the MTBS-NAs. The derivatization with PFBBr increased the retention time for the corresponding standards. This is a result of the increases in molecular weight of the compounds: 14 amu are added for each labile hydrogen in the NAs through methylation, whereas 114 amu are added to NAs using MTBSTFA, and 180 amu using PFBBr.

The mass spectral analysis of the NA standards mixture revealed several advantages of the PFBBr derivatization. All of the PFB derivative mass spectra show a fragment at 181m/z as the base peak or as the second most abundant ion, which corresponds to the pentafluorobenzyl radical fragment (Figs. 4i–l and S2 i–l). When the derivatization reaction is performed using a weak base, PFBBr reacts selectively with NAs [38,39], yielding a fragment ion at 181m/z. We used estradiol-17β as a negative control for the selective derivatization of carboxylic acids with PFBBr using a weak base. Estradiol-17β has two hydroxyl groups, one of them forming a phenolic moiety (hydroxyl 447

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

The signature fragment ion of PFB-NA standards at 181m/z is one of the 10 most abundant in the respective mass spectrum (Figs. 4i–l and S2i–l). The relative proportion of the fragment ion at 181m/z ranged from 33% to 99% (Table 4). It was predicted and determined that the commercial NA mixture mass spectra components also had the characteristic fragment ion at 181m/z. The variation in the proportion of the 181m/z ion is related to variations in the chemical stability of the derivatized NA to the electron impact ionization. For example, the mass spectrum of PFB-adamantane carboxylic acid has a characteristic fragment ion at 135m/z (in addition to the fragment ion at 181m/z) that corresponds to the polycyclic ring, which forms the core of the adamantanes (Figs. 4K and S2K). The observation that characteristic ions are conserved provides information that helps with the identification of such compounds (Figs. 4 and Fig. S2). In order to identify any compound using its mass spectrum data is necessary to obtain a “pure” mass spectrum, but when two or more components of a complex mixture cannot be resolved in the chromatographic phase, the mass spectrum of the unresolved peak contains the

group bonded to an aromatic hydrocarbon group). Using the reaction and conditions herein proposed for derivatization of NAs with PFBBr, the peak of estradiol-17β-PFB was not detected in the GC/EIMS chromatogram (data not shown), however, a peak of non-derivatized estradiol-17β was detected at 42.5 min. This is especially useful because the differentiation of carboxylic acids from other components in complex mixtures of NAs is possible with PFBBr as the derivatization reagent. It was not possible to identify a general characteristic fragment of the derivatized standards using BF3/MeOH or MTBSTFA. Nevertheless, MTBS-derivatives present a 75m/z ion corresponding to [C2H6OSi+H]+, which appears in one case as base peak (Figs. 4e–h and S2e–h). Another important advantage of using PFBBr is that the molecular ion [M]+ of all PFB derivatives can be detected. This was not possible in all of the cases using MTBSTFA ( Table 4). When MTBSTFA is used as derivatization reagent in the mixture of NA standards the molecular ion was not detected in 6 out of the 8 standards. In these 6 cases, the major fragmentations peaks of the MTBS derivatives were [M-CH3]+ due to the loss of a methyl group [47–49].

Fig. 5. GC-EIMS TIC of a Sigma NA mixture derivatized with BF3/MeOH (a), MTBSTFA (b) and PFBBr (c), shown in black. GC-EIMS TIC of a Merichem NA mixture derivatized with BF3/MeOH (d), MTBSTFA (e) and PFBBr (f), shown in black. GC-EIMS EIC (181m/z) from a Sigma and a Merichem NA mixture derivatized with PFBBr in c and f, shown in red. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

448

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

Fig. 5. (continued)

PFBBr. A similar trend was observed in the calculation of the limits of detection. The results are listed in Table 5. Overall, the derivatization with PFBBr provided the best results for the analysis of model NAs.

combined fragments of the coeluted components. In a complex NA mixture in which there is no information on the number of components and their respective molecular weights, there is a possibility that each fragment of an unresolved peak corresponds to the molecular weight of a compound. In the PFB derivatives, the possibility to assign an ion to the molecular weight of a component decreases considerably. The minimum expected molecular weight of the smallest PFB-NA is 268, whereas for methylated NAs is 102 and for silyl NAs is 202. Consequently, only an ion with a higher m/z than the molecular weight of the smallest derivatized NA can be the molecular ion. For MTBS derivatives, the loss of a methyl (m/z=15) was observed in some cases, therefore the expected molecular weight of the smallest MTBS-NA is 187, making the ion distribution analysis more difficult. The results obtained from the PFBBr derivatization lead to a more accurate NA identification and carboxylic acid characterization. In three out of the eight standards the limit of quantification was improved using MTBSTFA. In four out of the eight standards the limit of quantification was improved using PFBBr. The limit of quantification for 1-adamantanecarboxylic acid was the same using MTBSTFA and

3.4. Analysis of the oil extracted NA mixtures The three derivatization procedures were evaluated to study two complex oil extract NA mixtures (Sigma and Merichem). The time of reaction and molar ratio of derivatization reagent/NAs were as found to be optimal in the experiments with the NA standards. The reaction times for BF3/MeOH, MTBSTFA and PFBBr were 50, 18.5 and 10 min, respectively. The molar ratios of derivatization reagent/NAs for BF3/ MeOH, MTBSTFA and PFBBr were 1250, 25 and 49, respectively. From the analysis of a derivatized NA mixture with BF3/MeOH, it is not possible to obtain useful information for the identification of its components because the chromatogram shows a characteristic hump with unresolved peaks (Fig. 5a and d), similar to results that have been previously reported [13,22–26,39]. The chromatogram associated with the NA mixture derivatized with MTBSTFA (Fig. 5b and e) exhibits 449

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

Fig. 6. Ion distribution of a Sigma (a)–(c) and a Merichem (d)–(f) NA extract using BF3/MeOH (a) and (d), MTBSTFA (b) and (e) and PFBBr (c) and (f) as derivatization reagents. Values correspond to various carbon numbers and z families in the NA commercial mixtures. The color intensity represents the percentage, by number of ions of NAs in the mixture that account for a given carbon number in a given z family, corresponding to specific m/z values from GC–EIMS analysis. The sum of all the values equals 100%. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

MTBSTFA. A database was built to classify the probable molecular ions. The limitation of this analysis is that only ions that fit the classic formula of NAs can be included. The methylated NA, MTBS-NA and PFB-NA standards retention indices (Table 4) were used to obtain more accurate results of ion distribution. According to the retention indices (Ri), it is expected that components of the PFB-NA mixture elute at a specific retention time range according to the carbon number. According to the ion distribution of the methylated Sigma mixture, ~42% of the ions may be related to aliphatic NAs. Approximately 16% of the ions belongs to diaromatic NAs and the aliphatic NAS with 7 carbons being the most abundant (Fig. 6a). With the silyl ion distribution for the Sigma mixture, it was found that most of the ions are classified as aliphatic NAs (~36%), and among these NAs with 11 carbons were the most abundant group (Fig. 6b). However, using PFBBr as the derivatization reagent it was observed that ~75% of the compounds analyzed belong to the z-0 family (i.e., aliphatic naphthenic acids) with 4–18 carbons; among these, the NAs with 6 carbons were the most abundant (Fig. 6c). With the analysis of the Sigma NA mixture with the three reagents, it was observed that aliphatic NAs are the most abundant group of NAs in this mixture, confirming previous results using MTBSTFA in a Sigma extract [35]. The possible presence of NAs of the z-2 family, as well as aromatic compounds, was also observed. According to the ion distribution of the methylated Merichem mixture, ~24% of the ions may be related to z-4 NAs and monoaromatic NAs with 7 carbons being the most abundant (Fig. 6d). With the ion distribution of the silyl Merichem mixture, it was found that most of the ions are classified as z-6 NAs (~23%), and among these NAs with 12 carbons were the most abundant group (Fig. 6e). However, using PFBBr as derivatization reagent it was observed that ~33% of the compounds analyzed belong to the z-6 family with 12–19 carbons; among these, the NAs with 16 carbons were the most abundant (Fig. 6c). Correspondingly, the ion distributions depicted in Fig. 6c

better-resolved peaks than the methylation technique, but not to the same degree as with PFB derivatives (Fig. 5c and f). In an unknown sample it is important to obtain the molecular ions of the components, however using MTBSTFA in 6 out of the 8 standards the molecular ion was not detected, giving incomplete mass spectra. Therefore, for precise molecular weight determinations, a more complex spectral analysis is required. In the analysis of the Sigma mixture derivatized with PFBBr, 206 peaks were detected, of which 75% present the fragment ion at 181m/z. For the Merichem mixture 176 peaks were detected of which 68% present the fragment ion at 181m/z. As determined by a lack of the fragment ion at 181m/z several components in the mixtures were not derivatized. This was approximately 25% in the Sigma mixture and 32% in the Merichem mixture (Fig. 5c and f) so there are compounds in the mixtures that are not carboxylic acids since they do not react with PFBBr. The analysis of the peaks that did not have the fragment ion at 181m/z in the 2 oil extract mixtures showed “ski-slope” fragmentation patterns with peaks separated by 14m/z, which correspond to CH2 units. These mass spectra might correspond to branched or unbranched saturated hydrocarbons [51]. This outcome shows an important advantage of derivatizing complex NA mixtures with PFBBr. The detection of these non-derivatizable components has not been reported with the other derivatization methods, perhaps since complex mixtures of NAs have not yet been entirely resolved. In addition, the differentiation between derivatized and non-derivatized components in mixtures of NAs is not possible using BF3/MeOH and MTBSTFA. The mass spectrum of the peaks at the point of highest signal/noise ratio were analyzed to determine the ions that could be assigned to molecular weights of NAs. The peaks that did not have the fragment ion at 181m/z associated with PFFBr derivatization were discarded because these cannot be attributed to NAs. All of the detected peaks were used for the analysis of the mixtures using BF3/MeOH and 450

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

analysis of NA standards and mixtures yielded better results compared to using BF3/MeOH or MTBSTFA. The derivatization of NAs with PFBBr offers several advantages, including increased derivatization signal yields, resolution, and sensitivity. In the mass spectral analysis of PFB derivatives, a characteristic fragment from the ion pentafluorobenzyl was identified at 181m/z. In the analysis of the two oil extract NA mixtures derivatized with PFBBr, several peaks that lacked the fragment ion at 181m/z were detected; therefore, these peaks are not NAs. Together, our results indicate that PFBBr derivatization allows the differentiation of compounds with no labile hydrogens, greatly increasing the accuracy of analyzing complex NA mixtures. The use of PFBBr increases sensitivity, chromatographic resolution, and identification accuracy for the analysis of standards and mixtures of NAs.

and f for NA mixtures derivatized with PFBBr are different from those using the methylation and silylation techniques (Fig. 6a, b, d, and e) and also different from others previously reported [35,45]. The 2 oil NA extracts (Sigma and Merichem) were spiked with the 8 NA standards to confirm that PFBBr was reacting with all of the NAs present in the mixtures. The 8 NA standards were identified in the chromatogram, the mass spectra of the 8 NA standards derivatives were detected and the spectral analysis showed the presence of the molecular ion and the fragment ion at 181m/z in all of the standards extracted mass spectra (Figs. S3–S5). Based on the mass spectral analysis of the mixtures and the presumable presence of saturated hydrocarbons an open column chromatography approach was used to separate the components of the mixtures according to their polarity and to determine an approximate proportion of non-NA components. From the column, a hexane and a methanol fraction of each oil NA extract were obtained. The NAs are expected to be detected in the methanol fraction and the saturated hydrocarbons are expected to be detected in the hexane fraction. The components in the methanol fraction corresponded to 93% and 92% by weight of the Sigma and Merichem extracts, respectively. A sample of each of the fractions was derivatized with PFBBr. For the methanol fraction 96% and 94% of the peaks have the fragment ion at 181m/z of the Sigma and Merichem mixtures, respectively, therefore they are carboxylic acids. However, the fragment ion at 181m/z was not detected in the hexane fraction (Fig. S6). The components in the hexane fraction have “ski-slope” fragmentation patterns with peaks separated by 14m/z, which correspond to CH2 units, which is characteristic of branched or unbranched saturated hydrocarbons [51], so they might be present in both mixtures. For example, a peak detected at 26.9 min in the non-polar fraction of the Sigma extract (AMDIS, purity=50, s/n=68) was found to have a high match in the NIST Library with hexadecane (Match=840, R Match=840) (Fig. S7). The derivatization methodologies reported here focus on the analysis of carboxylic acids and their derivatives. An alternative and specific methodology would need to be developed for the analysis of trace non-carboxylic acid compounds in NA mixtures, such as non-volatile components, alcohols, ketones, and ethers that could also be present [13]. Moreover, low molecular weight alcohols, ketones, and ethers are very volatile and may be lost during sample processing. Thus, other components such as alcohols, ketones, and ethers, are not observed perhaps because they are not present in these specific mixtures, or they may be in low concentration compared to the carboxylic acids. The 2 mixtures (Sigma and Merichem) were spiked with the 8 standards to confirm that NAs were effectively separated by column chromatography in the methanol fraction. A sample of each of the two fractions per extract was taken and derivatized using PFBBr. The 8 standards were detected in the methanol fractions, confirming the effectiveness of this approach to separate NAs from other components. The 8 exhibited the characteristic fragment ion at 181m/z as expected. The standards were not detected in the hexane fraction, nor was the fragment ion at 181m/z (Fig. S8). On average the open column recovery was 97 ± 2%. Sigma and Merichem mixtures have been used as reference mixtures for validation, calibration and quantification of NAs in water samples assuming a high concentration of NAs [16,52,53]. Our results showed that these standards contain ~7% by weight of other components. Herein, we have found that separation by column chromatography in combination with derivatization using PFBBr, offers a more accurate characterization of carboxylic acids in NA mixtures. Significantly, these non-NA components in the mixtures were evident because they did not contain the fragment ion at 181m/z that resulted from the use of the PFBBr derivatization reagent.

Acknowledgements Grant support for JMGV was provided by CONACyT Mexico. The support of NSERC Strategic Grants Program (RGPIN2013), University of Ottawa Research Chair program and Environment Canada for VLT are also acknowledged with appreciation. We would like to thank Professor John T. Arnason, Ammar Saleem, Julie Bilodeau, Jesse Leonard and Erin Blake for proofreading and providing helpful feedback that improved the manuscript. Helpful discussions with Professor Jules M. Blais are acknowledged. The gifts of the Merichem mixture from Dr. John Headley, and 1-adamantane acetic acid and 3-methyl-1adamantane acetic acid from Steven J. Rowland are acknowledged with appreciation. Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.talanta.2016.10.057. References [1] A.D. McNaught, A. Wilkinson, IUPAC. Compendium of Chemical Terminology (The “Gold Book”), Blackwell Scientific Publications, Oxford, 1997. [2] J.A. Brient, P.J. Wessner, M.N. Doyle, Naphthenic Acids, Kirk-Othmer Encyclopedia of Chemical Technology, John Wiley & Sons, Inc., Hoboken, New Jersey, USA, 2000. [3] X.Q. Wu, H.M. Jing, Y.G. Zheng, Z.M. Yao, W. Ke, Resistance of Mo-bearing stainless steels and Mo-bearing stainless-steel coating to naphthenic acid corrosion and erosion–corrosion, Corros. Sci. 46 (4) (2004) 1013–1032. [4] X. Wu, H. Jing, Y. Zheng, Z. Yao, W. Ke, Erosion–corrosion of various oil-refining materials in naphthenic acid, Wear 256 (1–2) (2004) 133–144. [5] R. Kane, M. Cayard, A comprehensive study of naphthenic acid corrosion, Corrosion, 2002 [6] C. Aeppli, C.A. Carmichael, R.K. Nelson, K.L. Lemkau, W.M. Graham, M.C. Redmond, D.L. Valentine, C.M. Reddy, Oil weathering after the deepwater horizon disaster led to the formation of oxygenated residues, Environ. Sci. Technol. 46 (16) (2012) 8799–8807. [7] M.K. McNutt, R. Camilli, T.J. Crone, G.D. Guthrie, P.A. Hsieh, T.B. Ryerson, O. Savas, F. Shaffer, Review of flow rate estimates of the deepwater horizon oil spill, Proc. Natl. Acad. Sci. USA 109 (50) (2012) 20260–20267. [8] Y. Wan, B. Wang, J.S. Khim, S. Hong, W.J. Shim, J. Hu, Naphthenic acids in coastal sediments after the Hebei Spirit oil spill: a potential indicator for oil contamination, Environ. Sci. Technol. 48 (7) (2014) 4153–4162. [9] M.D. MacKinnon, H. Boerger, Description of two treatment methods for detoxifying oil sands tailings pond water, Water Qual. Res. J. Can. 21 (4) (1986) 496–512. [10] S.D. Richardson, T.A. Ternes, Water analysis: emerging contaminants and current issues, Anal. Chem. 86 (6) (2014) 2813–2848. [11] M.P. Barrow, J.V. Headley, K.M. Peru, P.J. Derrick, Data visualization for the characterization of naphthenic acids within petroleum samples, Energy Fuels 23 (5) (2009) 2592–2599. [12] M.P. Barrow, M. Witt, J.V. Headley, K.M. Peru, Athabasca oil sands process water: characterization by atmospheric pressure photoionization and electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry, Anal. Chem. 82 (9) (2010) 3727–3735. [13] X. Ortiz, K.J. Jobst, E.J. Reiner, S.M. Backus, K.M. Peru, D.W. McMartin, G. O’Sullivan, V.Y. Taguchi, J.V. Headley, Characterization of naphthenic acids by gas chromatography-Fourier transform ion cyclotron resonance mass spectrometry, Anal. Chem. 86 (15) (2014) 7666–7673. [14] X. Wang, K.L. Kasperski, Analysis of naphthenic acids in aqueous solution using HPLC-MS/MS, Anal. Methods 2 (11) (2010) 1715–1722. [15] R. Hindle, M. Noestheden, K. Peru, J. Headley, Quantitative analysis of naphthenic

4. Conclusions This study reports on the effectiveness of PFBBr as a derivatization reagent for the analysis of NAs. Using PFBBr prior to GC-EIMS 451

Talanta 162 (2017) 440–452

J.M. Gutierrez-Villagomez et al.

[16]

[17]

[18]

[19]

[20] [21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32] [33]

[34]

[35]

acids in water by liquid chromatography–accurate mass time-of-flight mass spectrometry, J. Chromatogr. A 1286 (2013) 166–174. P. Brunswick, D. Shang, G. van Aggelen, R. Hindle, L.M. Hewitt, R.A. Frank, M. Haberl, M. Kim, Trace analysis of total naphthenic acids in aqueous environmental matrices by liquid chromatography/mass spectrometry-quadrupole time of flight mass spectrometry direct injection, J. Chromatogr. A 1405 (2015) 49–71. K. Duncan, D. Volmer, C. Gill, E. Krogh, Rapid screening of carboxylic acids from waste and surface waters by ESI-MS/MS using barium ion chemistry and on-line membrane sampling, J. Am. Soc. Mass Spectrom. (2015) 1–8. R. Huang, K.N. McPhedran, M. Gamal El-Din, Ultra performance liquid chromatography ion mobility time-of-flight mass spectrometry characterization of naphthenic acids species from oil sands process-affected water, Environ. Sci. Technol. 49 (19) (2015) 11737–11745. A.S. Pereira, J.W. Martin, Exploring the complexity of oil sands process-affected water by high efficiency supercritical fluid chromatography/orbitrap mass spectrometry, Rapid Commun. Mass Spectrom.: RCM 29 (8) (2015) 735–744. W.W. Christie, Why I dislike boron trifluoride-methanol, Lipid Technol. 6 (1994) 66–68. A.G. Shepherd, V. van Mispelaar, J. Nowlin, W. Genuit, M. Grutters, Analysis of naphthenic acids and derivatization agents using two-dimensional gas chromatography and mass spectrometry: impact on flow assurance predictions, Energy Fuels 24 (4) (2010) 2300–2311. D.M. Jones, J.S. Watson, W. Meredith, M. Chen, B. Bennett, Determination of naphthenic acids in crude oils using nonaqueous ion exchange solid-phase extraction, Anal. Chem. 73 (3) (2001) 703–707. S.J. Rowland, A.G. Scarlett, D. Jones, C.E. West, R.A. Frank, Diamonds in the rough: identification of individual naphthenic acids in oil sands process water, Environ. Sci. Technol. 45 (7) (2011) 3154–3159. S.J. Rowland, C.E. West, D. Jones, A.G. Scarlett, R.A. Frank, L.M. Hewitt, Steroidal aromatic ‘naphthenic acids' in oil sands process-affected water: structural comparisons with environmental estrogens, Environ. Sci. Technol. 45 (22) (2011) 9806–9815. S.J. Rowland, C.E. West, A.G. Scarlett, D. Jones, R.A. Frank, Identification of individual tetra- and pentacyclic naphthenic acids in oil sands process water by comprehensive two-dimensional gas chromatography/mass spectrometry, Rapid Commun. Mass Spectrom.: RCM 25 (9) (2011) 1198–1204. D. Jones, C.E. West, A.G. Scarlett, R.A. Frank, S.J. Rowland, Isolation and estimation of the ‘aromatic’ naphthenic acid content of an oil sands process-affected water extract, J. Chromatogr. A 1247 (2012) 171–175. C.E. West, A.G. Scarlett, A. Tonkin, D. O’Carroll-Fitzpatrick, J. Pureveen, E. Tegelaar, R. Gieleciak, D. Hager, K. Petersen, K.E. Tollefsen, S.J. Rowland, Diaromatic sulphur-containing ‘naphthenic’ acids in process waters, Water Res. 51 (2014) 206–215. M.J. Wilde, C.E. West, A.G. Scarlett, D. Jones, R.A. Frank, L.M. Hewitt, S.J. Rowland, Bicyclic naphthenic acids in oil sands process water: identification by comprehensive multidimensional gas chromatography–mass spectrometry, J. Chromatogr. A 1378 (2015) 74–87. D.R. Parkinson, 2.26 – Analytical derivatization techniques, in: J. Pawliszyn (Ed.) Comprehensive Sampling and Sample Preparation, Academic Press, Oxford, 2012, pp. 559–595. W.P. John St., J. Rughani, S.A. Green, G.D. McGinnis, Analysis and characterization of naphthenic acids by gas chromatography–electron impact mass spectrometry of tert.-butyldimethylsilyl derivatives, J. Chromatogr. A 807 (2) (1998) 241–251. F.M. Holowenko, M.D. MacKinnon, P.M. Fedorak, Characterization of naphthenic acids in oil sands wastewaters by gas chromatography-mass spectrometry, Water Res. 36 (11) (2002) 2843–2855. T.E. Havre, J. Sjöblom, J.E. Vindstad, Oil/water-partitioning and interfacial behavior of naphthenic acids, J. Dispers. Sci. Technol. 24 (6) (2003) 789–801. M. Merlin, S.E. Guigard, P.M. Fedorak, Detecting naphthenic acids in waters by gas chromatography-mass spectrometry, J. Chromatogr. A 1140 (1–2) (2007) 225–229. A.C. Scott, R.F. Young, P.M. Fedorak, Comparison of GC–MS and FTIR methods for quantifying naphthenic acids in water samples, Chemosphere 73 (8) (2008) 1258–1264. F.C. Damasceno, L.D.A. Gruber, A.M. Geller, M.C. Vaz de Campos, A.O. Gomes,

[36]

[37]

[38] [39]

[40]

[41]

[42]

[43]

[44] [45]

[46] [47]

[48]

[49]

[50]

[51]

[52]

[53]

452

R.C.L. Guimaraes, V.F. Peres, R.A. Jacques, E.B. Caramao, Characterization of naphthenic acids using mass spectroscopy and chromatographic techniques: study of technical mixtures, Anal. Methods 6 (3) (2014) 807–816. F.K. Kawahara, Microdetermination of derivatives of phenols and mercaptans by means of electron capture gas chromatography, Anal. Chem. 40 (6) (1968) 1009–1010. F. Orata, Derivatization reactions and reagents for gas chromatography analysis, in: M. Ali Mohd (Ed.)Advanced Gas Chromatography - Progress in Agricultural, Biomedical and Industrial Applications, INTECH Open Access Publisher, Rijeka, Croatia, 2012. Pentafluorobenzyl Bromide, Hexaoxacyclooctadecane, Product Specification, in: Supelco (Ed.), Sigma B. Davis, Crown ether catalyzed derivatization of carboxylic acids and phenols with pentafluorobenzyl bromide for electron capture gas chromatography, Anal. Chem. 49 (6) (1977) 832–834. C.-J. Chien, M.J. Charles, K.G. Sexton, H.E. Jeffries, Analysis of airborne carboxylic acids and phenols as their pentafluorobenzyl derivatives: gas chromatography/ion trap mass spectrometry with a novel chemical ionization reagent, PFBOH, Environ. Sci. Technol. 32 (2) (1998) 299–309. E. Gracia-Moreno, R. Lopez, V. Ferreira, Quantitative determination of five hydroxy acids, precursors of relevant wine aroma compounds in wine and other alcoholic beverages, Anal. Bioanal. Chem. 407 (26) (2015) 7925–7934. O. Quehenberger, A.M. Armando, E.A. Dennis, High sensitivity quantitative lipidomics analysis of fatty acids in biological samples by gas chromatography-mass spectrometry, Biochim. Biophys. Acta 1811 (11) (2011) 648–656. M.A. Bezerra, R.E. Santelli, E.P. Oliveira, L.S. Villar, L.A. Escaleira, Response surface methodology (RSM) as a tool for optimization in analytical chemistry, Talanta 76 (5) (2008) 965–977. G. Sun, P. Stremple, Retention index, Characterization of Flavor, Fragance and Many Other Compounds on DB1 and DB-XLB, J & W Scientific J.S. Clemente, N.G. Prasad, M.D. MacKinnon, P.M. Fedorak, A statistical comparison of naphthenic acids characterized by gas chromatography-mass spectrometry, Chemosphere 50 (10) (2003) 1265–1274. D.R. Knapp, Handbook of Analytical Derivatization Reactions, Wiley, Hoboken, New Jersey, USA, 1979. J. Wu, R. Hu, J. Yue, Z. Yang, L. Zhang, Study on the derivatization process using NO-bis-(trimethylsilyl)-trifluoroacetamide, N-(tert-butyldimethylsilyl)-N-methyltrifluoroacetamide, trimethylsilydiazomethane for the determination of fecal sterols by gas chromatography-mass spectrometry, World Acad. Sci. Eng. Technol. Int. J. Chem. Mol. Nucl. Mater. Metall. Eng. 2 (8) (2008) 175–178. A. Shareef, C.J. Parnis, M.J. Angove, J.D. Wells, B.B. Johnson, Suitability of N, Obis (trimethylsilyl) trifluoroacetamide and N-(tert-butyldimethylsilyl)-N-methyltrifluoroacetamide as derivatization reagents for the determination of the estrogens estrone and 17α-ethinylestradiol by gas chromatography–mass spectrometry, J. Chromatogr. A 1026 (1) (2004) 295–300. A. Shareef, M.J. Angove, J.D. Wells, Optimization of silylation using N-methyl-N(trimethylsilyl)-trifluoroacetamide, N, O-bis-(trimethylsilyl)-trifluoroacetamide and N-(tert-butyldimethylsilyl)-N-methyltrifluoroacetamide for the determination of the estrogens estrone and 17α-ethinylestradiol by gas chromatography–mass spectrometry, J. Chromatogr. A 1108 (1) (2006) 121–128. J.I. Cacho, N. Campillo, P. Viñas, M. Hernández-Córdoba, Evaluation of three headspace sorptive extraction coatings for the determination of volatile terpenes in honey using gas chromatography–mass spectrometry, J. Chromatogr. A 1399 (2015) 18–24. J.T. Watson, O.D. Sparkman, Electron Ionization, Introduction to Mass Spectrometry, John Wiley & Sons, Ltd, Chichester, West Sussex, England, 2008, pp. 315–448. M.B. Woudneh, M. Coreen Hamilton, J.P. Benskin, G. Wang, P. McEachern, J.R. Cosgrove, A novel derivatization-based liquid chromatography tandem mass spectrometry method for quantitative characterization of naphthenic acid isomer profiles in environmental waters, J. Chromatogr. A 1293 (2013) 36–43. W. Lu, A. Ewanchuk, L. Perez-Estrada, D. Sego, A. Ulrich, Limitation of fluorescence spectrophotometry in the measurement of naphthenic acids in oil sands process water, J. Environ. Sci. Health Part A Toxic/Hazard. Subst. Environ. Eng. 48 (4) (2013) 429–436.